Detecting selection using extended haplotype homozygosity (EHH)-based statistics in unphased or unpolarized data

Area of Science:

• Population genetics
• Evolutionary biology
• Genomics

Background:

• Detecting positive selection in genomic data is vital for understanding evolution.
• Extended haplotype homozygosity (EHH) statistics are commonly used but often require phased haplotypes and polarized variants.
• Existing methods have limitations in flexibility and data requirements.

Purpose of the Study:

• To unify and extend modifications to EHH statistics, relaxing requirements for phased haplotypes and polarized variants.
• To evaluate the performance of modified statistics against original versions using simulations and empirical data.
• To assess the impact of phasing and ancestry information on the accuracy of selection detection.

Main Methods:

• Development of unified and extended EHH statistics.
• Simulation studies to measure the false discovery rate in whole-genome scans.
• Analysis of empirical population genetic data to quantify overlap in candidate regions.
• Comparison of modified statistics with original versions.

Main Results:

• Phasing information is indispensable for accurate within-population statistics, except in very large samples.
• Phasing information is also crucial for cross-population statistics in small samples.
• Ancestry information has a relatively minor impact on both types of statistics.
• Modified statistics demonstrate comparable or improved performance in simulations and empirical data.

Conclusions:

• The developed modifications enhance the flexibility and accuracy of detecting positive selection using EHH statistics.
• Phasing is a critical factor for reliable population genetic analyses, particularly for within-population and small-sample cross-population comparisons.
• The R package 'rehh' is updated to include these novel statistics, facilitating their application in research.